I-12_Trend Analysis Summary Sheet EN8.9ER May2019

CITY OF DURHAM STORMWATER & GIS SERVICES
TREND ANALYSIS SUMMARY SHEET
MAY 13, 2019
Watershed: Eno River
Trend
Fecal Coliform, Turbidity, Hardness, Specific Conductivity, and Water Quality Index (WQI)
parameters:
Data Period: 1/1/2004 to 12/31/2018
Regulatory Status:
303(d) Listed TMDL Watershed Falls Lake ✓ Jordan Lake
Primary Data Sources
Secondary Data Sources
L
Durham Site Number
EN8.9ER
Upper Cape Fear Site
Number
NA
Durham Site Name
Eno River at Roxboro Rd
(West Pt on the Eno)
Upper Cape Fear Site Name
NA
USGS Discharge Site Number
02085070
Upper Neuse Site Number
ENR-16
USGS Discharge Site Name
Eno River nr Durham, NC
Upper Neuse Site Name
Eno River —at US 15 501
DWR—STORET Site Number
J0770000
DWR—STORET Site Name
Eno River at US 501 nr
Durham
Drainage Area (ac)
90,240
NDWRF Site Number
NA
Drainage Area (sqm)
141
NDWRF Site Name
NA
Comments:
Trend analysis was not performed for several typical water quality parameters because the percent of values less than
a detection limit (nondetects) were much higher than 5%, which is the recommended threshold for the Kendall trend
tests.' The following parameters were not analyzed for trends due to the high percentage of nondetects and varying
detection limits: dissolved copper, dissolved zinc, biochemical oxygen demand, ammonia nitrogen, nitrate+nitrite, total
kjeldahl nitrogen, total phosphorus, and total suspended solids. Therefore, trends were analyzed for parameters
having less than 5% nondetects (%NDs), which are fecal coliform, turbidity, hardness, specific conductivity, and the
City's Water Quality Index. Data were downloaded online from the City's Water Quality Data Web Portal for
monitoring site EN8.9ER and the National Water Quality Web Portal for NCDWR's monitoring site J0770000. Both
monitoring sites are located at the Eno River near the US 15-501 bridge crossing (also near West Point on the Eno). The
data period from 1/1/2004 to 12/31/2018 represents 15 calendar years and 16 water years (e.g., October 1, 2004 —
September 30, 2005). Duplicate results were replaced with an average of the two values. Instantaneous and daily
discharge measurements were downloaded online from USGS for site 02085070, which is located near the City and
State monitoring site. After internal discussions with the Water Quality Manager (M. Woolfolk), the decision was made
to only use daily average discharge measurements for the Kendall Trend Tests (Mann -Kendall and Seasonal Kendall).
Summary statistics were calculated using the EPA's ProUCL software (Version 5.1) because of its ability to handle data
' City of Durham. 2015. City of Durham Water Quality Trend Analysis Standard Operating Procedures. Public Works Department,
Stormwater & GIS Services Division. March 30, 2015. Durham, NC.

sets with and without nondetects and data sets that have multiple detection limits. ProUCL estimates summary
statistics for censored data by using the nonparametric Kaplan -Meier method .Z Summary statistics for detected data
were also prepared using ProUCL.
The nonparametric Mann -Kendall (MK) and Seasonal Kendall (SK) trend tests (USGS, 2009 Version) were performed
with and without LOWESS (locally weighted scatterplot smoothing). The relationship between concentration data and
flow data were analyzed using the MK and SK LOWESS tests. As recommended in the Water Quality Trend Analysis SOP,
the Kendall trend tests should be performed using different smoothing windows (values off between 0 and 1) to
compare the p-values and their significance. The Kendall trend tests can be used for the detection of trend in censored
data sets, but all censored and uncensored values less than or equal to the highest reporting limit must be set equal to
the detection limit3. Due to varying detection limits during the trend analysis period, several parameters were not
analyzed for a trend because resetting or correcting the data increased the number of censored data.
The alpha level (a) was set at 0.05 which means that the trend is statistically significant at 95% or greater confidence
level. The Seasonal Kendall trend test literature suggests using the adjusted p-value when there are more than 10 years
of data .4 The only significant trend identified at EN8.9ER was a downward trend for specific conductivity. The Seasonal
Kendall without LOWESS (concentration data) test resulted in a p-value of 0.05, an adjusted p-value of 0.0541, and a
Kendall test statistic (S) of -136. Although the Seasonal Kendall without LOWESS test for fecal coliform data resulted in
a p-value of 0.0082, the adjusted p-value of 0.0663 is not significant and falls outside the 95% confidence level. The
Kendall test statistic (S) for fecal coliform is 184 which would indicate an upward trend if it were considered significant.
The Seasonal Kendall trend lines are included on the concentration vs. date graphs for specific conductivity and fecal
coliform. Graphs were created using JMP software (version 13). Based on the results of the Kendall trend tests, no
other significant trends were identified.
Abbreviations
Abbreviation
Name
%ND
percent of nondetects
adjusted p
adjusted p-value
cfu
colony forming units
f
smoothing function
KM
Kaplan -Meier Method
LOWESS
locally weighted scatterplot smoothing
MDL
method detection limit
mg/L
milligrams per liter
MK
Mann -Kendall
N
number of observations
ND
nondetects
NTU
Nephelometric turbidity units
Num Detects
Number of detected values
PQL
practical quantitation limit
p-value
probability value or significance
S
Kendall test statistic
SK
Seasonal Kendall
Z United States Environmental Protection Agency. 2015. ProUCL Version 5.1.002 User Guide: Statistical Software for Environmental
Applications for Data Sets with and without Nondetect Observations. EPA/600/R-07/041. USEPA Office of Research and
Development. Washington, DC.
3 Hirsch, R.M., Alexander, R.B., and Smith, R.A. 1991. Selection of methods for the detection and estimation of trends in water
quality: Water Resources Research v. 27, p. 803-813.
4 Hirsch, R.M., and Slack, J.R. 1984. A nonparametric trend test for seasonal data with serial dependence. Water Resources Research
v. 20, p. 727-732.

a I alpha level
Censored data (nondetects and detects) summary table
Parameter
Units
N
Num Detects
Num NDs
% NDs
Min ND
Max ND
Ammonia Nitrogen
mg/L
344
140
204
59%
0.01
0.09
Biochemical Oxygen Demand
mg/L
177
20
157
89%
2
2.6
Conductivity
µS/cm
358
358
0
0%
N/A
N/A
Dissolved Copper
µg/L
177
82
95
54%
5
5
Dissolved Oxygen
mg/L
357
357
0
0%
N/A
N/A
Dissolved Oxygen Saturation
%
177
177
0
0%
N/A
N/A
Fecal Coliform
cfu/100ml-
316
309
7
2%
1
20
Hardness
mg/L
177
177
0
0%
N/A
N/A
Nitrate + Nitrite as N
mg/L
343
287
56
16%
0.01
0.1
Total Kjeldahl Nitrogen
mg/L
339
259
80
24%
0.03
0.8
Total Phosphorus
mg/L
339
256
83
24%
0.01
0.05
Total Suspended Solids
mg/L
289
211
78
27%
1
13
Turbidity
NTU
321
321
0
0%
N/A
N/A
Dissolved Zinc
µg/L
159
80
79
50%
10
10
Detection limits and corrected censored values for Kendall Trend Tests
City of Durham
Detection Limits (MDL)
NCDWR Detection Limits (PQL)
Parameter
Min
Max
Mode
Corrected
Value
Min
Max
Mode
Corrected
Value
Ammonia Nitrogen (mg/L)
0.01
0.09
0.05
0.05
0.02
0.02
0.02
0.05
Biochemical Oxygen Demand (mg/L)
2
2.6
2
2
n/a
n/a
n/a
n/a
Copper(µg/L)
0.04
5
5
5
n/a
n/a
n/a
n/a
Fecal Coliform (cfu/100ml)
1
20
20
20
1
1
1
20
Nitrate + Nitrite as N (mg/L)
0.008
0.1
0.1
0.1
0.02
0.02
0.02
0.1
Total Kjeldahl Nitrogen (mg/L)
0.03
0.8
0.5
0.5
0.2
0.2
0.2
0.5
Total Phosphorus (mg/L)
0.01
0.05
0.03
0.05
0.02
0.02
0.02
0.05
Total Suspended Solids (mg/L)
1
6.3
2.5
6.2
2.5
13
6.2
6.2
Zinc (µg/L)
0.7
10
10
10
n/a
n/a
n/a
n/a
Summary statistics for censored data set (with nondetects) using the Kaplan -Meier (KM) Method
Parameter
Units
N
KM Mean
KM
Variance
KM Standard
Deviation
KM Coefficient of
Variation
Ammonia Nitrogen
mg/L
344
0.02
0.0003
0.018
0.76
Biochemical Oxygen Demand
mg/L
177
2.1
0.09
0.30
0.15
Conductivity
[IS/CM
358
121
1065
33
0.27
Dissolved Copper
µg/L
177
1.64
0.19
0.43
0.27
Dissolved Oxygen
mg/L
357
9.61
6.40
2.53
0.26

Dissolved Oxygen Saturation
%
177
95.3
65.1
8.1
0.08
Fecal Coliform
cfu/100mL
316
481
5683388
2384
4.95
Hardness
mg/L
177
34
98
10
0.29
Nitrate + Nitrite as N
mg/L
343
0.24
0.04
0.19
0.81
Total Kjeldahl Nitrogen
mg/L
339
0.36
0.03
0.16
0.45
Total Phosphorus
mg/L
339
0.04
0.001
0.03
0.81
Total Suspended Solids
mg/L
289
9.3
471
21.7
2.33
Turbidity
NTU
321
12.8
613
24.8
1.93
Dissolved Zinc
µg/L
159
4.71
27.7
5.27
1.12
Summary statistics for detected data set only
Standard
Coefficient
Parameter
Units
N
Min
Max
Mean
Median
Variance
Skewness
Deviation
of Variation
Ammonia Nitrogen
mg/L
140
0.01
0.11
0.04
0.03
0.0004
0.02
1.5
0.54
Biochemical
mg/L
20
2
4
2.5
2
0.57
0.75
0.98
0.30
Oxygen Demand
Conductivity
PS/cm
358
67
293
121
114
1065
33
1.4
0.27
Dissolved Copper
µg/L
82
0.5
2.8
1.6
1.6
0.19
0.44
0.19
0.27
Dissolved Oxygen
mg/L
357
4.1
18.1
9.6
9.2
6.4
2.5
0.46
0.26
Dissolved Oxygen
%
177
72.6
120
95.3
95.1
65.1
8.07
0.115
0.08
Saturation
Fecal Coliform
cfu/100mL
309
2
34000
492
58
5825738
2414
10
5
Hardness
mg/L
177
10
80
34
32
98
10
0.80
0.29
Nitrate + Nitrite as
mg/L
287
0.01
1.3
0.28
0.26
0.03
0.18
1.42
0.66
N
Total Kjeldahl
mg/L
259
0.12
1.65
0.40
0.38
0.02
0.16
2.63
0.40
Nitrogen
Total Phosphorus
mg/L
256
0.02
0.42
0.05
0.04
0.001
0.04
5.29
0.73
Total Suspended
mg/L
211
1
275
12
6
623
25
7.30
2.11
Solids
Turbidity
NTU
321
1.67
364
12.8
6.53
613
24.8
9.87
1.93
Dissolved Zinc
µg/L
80
0.45
43
5.9
3.6
48
6.9
3.1
1.2
Trend tests: Mann -Kendall Test for Trend and Seasonal Kendall Test for Trend (USGS, 2009 Version)
"The Seasonal Kendall test performs the Mann -Kendall (MK) trend test for individual seasons of the year, where season
is defined by the user. It then combines the individual results into one overall test for whether the dependent (Y)
variable changes in a consistent direction (monotonic trend) over time. The Mann Kendall test computes Kendall's tau
nonparametric correlation coefficient and its test of significance for any pair of X,Y data.i5
5 Helsel, D.R., Mueller, D.K., and Slack, J.R. 2006. Computer program for the Kendall family of trend tests: U.S. Geological Survey
Scientific Investigations Report 2005-5275, 4 p.

Conductivity
Trend Summary Table - Conductivity
Analysis
S
Trend
p-value
Adjusted p
Significant
(a = 0.05)
Mann -Kendall, no LOWESS
-2965
Down
0.1901
n/a
No
Mann -Kendall, LOWESS, f = 0.2
-2438
Down
0.2814
n/a
No
Mann -Kendall, LOWESS, f = 0.5
-3264
Down
0.1493
n/a
No
Mann -Kendall, LOWESS, f = 0.8
-3300
Down
0.1448
n/a
No
Seasonal Kendall, no LOWESS
-136
Down
0.05
0.0541
Yes
Seasonal Kendall, LOWESS, f = 0.2
-78
Down
0.2639
0.2441
No
Seasonal Kendall, LOWESS, f = 0.5
-72
Down
0.303
0.2988
No
Seasonal Kendall, LOWESS, f = 0.8
-62
Down
0.3761
0.3972
No
Graph of Conductivity (µS/cm) vs Date
Line of fit equation: Y = 209.2 — 2.67e-8 * Date
Seasonal Kendall (no LOWESS) trend equation: Y = 121.2 +-0.8402 * Time
where Time = Year (as a decimal) — 2003.75
300
—Line cf Fit
— Sea scnaI Kendall trend line
250 ■
•
E 200
go
i
C • • r ti
• • •
■ LL .}� . .• • • ■ ■ •
• •tip ■ • • • • ■ • t • ■
•
f •
•
• • r
a
50
0112002 0112004 0112006 0112008 0112010 0112012 0112014 0112016 0112018 0112020
Date

Fecal Coliform
Trend Summary Table — Fecal Coliform
Analysis
S
Trend
p-value
Adjusted p
Significant
(a = 0.05)
Mann -Kendall, no LOWESS
-366
Down
0.644
n/a
No
Mann -Kendall, LOWESS, f = 0.2
-126
Down
0.875
n/a
No
Mann -Kendall, LOWESS, f = 0.5
-12
Down
0.989
n/a
No
Mann -Kendall, LOWESS, f = 0.8
-206
Down
0.7965
n/a
No
Seasonal Kendall, no LOWESS
184
Up
0.0082
0.0663
No*
Seasonal Kendall, LOWESS, f = 0.2
78
Up
0.2677
0.3579
No
Seasonal Kendall, LOWESS, f = 0.5
104
Up
0.2108
0.2108
No
Seasonal Kendall, LOWESS, f = 0.8
140
Up
0.0454
0.0885
No
*The trend is not significant when the a level is compared to the adjusted p-value.
40, '�
30000
20000
100M00
6000
5000
40M
3"
2000
1000pp
E 600
p 500
400
300
E 200
`o
v 100
V
60
d 50
LL 40
30
Graph of Fecal Coliform (cfu/100mL) vs Date (log scale)
Line of fit equation: Y = 1176-2.048e-7 * Date
Seasonal Kendall (no LOWESS) trend equation: Y = 59.64 +3.045 * Time
where Time = Year (as a decimal) — 2003.75
—Line cf Fit
Seascnal Kendall trend line
r ■ • • • •
5 •
4 • • •
3 ■
2
1 ■ r
0.8
01/2002 01/2004 01/2006 0112008 0112010 0112012 01/2014 0112016 01/2018 0112020
Date

Hardness
Trend Summary Table — Hardness
Analysis
S
Trend
p-value
Adjusted p
Significant
(a = 0.05)
Mann -Kendall, no LOWESS
629
Up
0.4206
n/a
No
Mann -Kendall, LOWESS, f = 0.2
1164
Up
0.1401
n/a
No
Mann -Kendall, LOWESS, f = 0.5
1096
Up
0.1648
n/a
No
Mann -Kendall, LOWESS, f = 0.8
956
Up
0.2257
n/a
No
Seasonal Kendall, no LOWESS
20
Up
0.7768
0.8639
No
Seasonal Kendall, LOWESS, f = 0.2
88
Up
0.2033
0.4082
No
Seasonal Kendall, LOWESS, f = 0.5
88
Up
0.2033
0.4304
No
Seasonal Kendall, LOWESS, f = 0.8
76
Up
0.2727
0.5097
No
Graph of Hardness (mg/L) vs Date
Line of fit equation: Y = 27.32 + 1.965e-9 * Date
0112002 01/2DD4 01120D6 0112008 01/2010 0112012 0112014 01j2016 01/2018 01/202D
Date

Turbidity
Trend Summary Table - Turbidity
Analysis
S
Trend
p-value
Adjusted p
Significant
(a = 0.05)
Mann -Kendall, no LOWESS
2073
Up
0.2854
n/a
No
Mann -Kendall, LOWESS, f = 0.2
888
Up
0.6474
n/a
No
Mann -Kendall, LOWESS, f = 0.5
302
Up
0.8767
n/a
No
Mann -Kendall, LOWESS, f = 0.8
-82
Down
0.9667
n/a
No
Seasonal Kendall, no LOWESS
49
Up
0.486
0.5542
No
Seasonal Kendall, LOWESS, f = 0.2
6
Up
0.9422
0.9552
No
Seasonal Kendall, LOWESS, f = 0.5
-12
Down
0.8732
0.9036
No
Seasonal Kendall, LOWESS, f = 0.8
-32
Down
0.6529
0.7333
No
Graph of Turbidity (NTU) vs Date (log scale)
Line of fit equation: Y = -18.17 + 9.128e-9 * Date
■
—Line of Fit
ao
70 ■
60 +
50
4a
3a
� a • • f
20 ti • r ■ • •
Oo
10 • .. r +.
7-
io •
6 .■ • • ■ i• ■ ■y •. • • • .�.
5 • • • • •• +' M • ■ • •y
• • • •• • 1• • i • • •
3 • • rr+ ; r •' + r
• • r ••
1
U.S
0.7
M
0.5
M
03
02
0.1
0112002 0112004 01j2006 01/20M 0112010 0112012 01/2014 0112016 0112018 0112020
Date

Water Quality Index (WQI)
Trend Summary Table — WQI
Analysis
S
Trend
p-value
Adjusted p
Significant
(a = 0.05)
Mann -Kendall, no LOWESS
-366
Down
0.644
n/a
No
Mann -Kendall, LOWESS, f = 0.2
-126
Down
0.875
n/a
No
Mann -Kendall, LOWESS, f = 0.5
-12
Down
0.989
n/a
No
Mann -Kendall, LOWESS, f = 0.8
-206
Down
0.7965
n/a
No
Seasonal Kendall, no LOWESS
-33
Down
0.6352
0.7333
No
Seasonal Kendall, LOWESS, f = 0.2
28
Up
0.6953
0.8042
No
Seasonal Kendall, LOWESS, f = 0.5
30
Up
0.6739
0.7921
No
Seasonal Kendall, LOWESS, f = 0.8
26
Up
0.7168
0.8252
No
Graph of WQI vs Date
Line of fit equation: Y = 97.12 - 1.288e-9 * Date
100
90
80
WQI
7D
6D
5D I I I '
D1/D1/2DD2 D1/D1/2D04 01/D1/2DD6 D1/D1/2DD8 D1/D1/2D1D D1/D1/2D12 D1/D1/2D14 D1/D1/2D16 D1/D1/2D18 D1/D1/2D2fl
Date